1. Field of the Invention
The present invention relates generally to a semiconductor device and particularly to a high speed semiconductor device having a SiGeC ternary mixed crystal semiconductor layer.
2. Description of the Related Art
Today, the silicon (Si) bipolar transistor is a classic semiconductor device. In the conventional silicon (Si) bipolar transistor, the carrier mobility within silicon is very limited which in turn limits the operational speed of the device. Therefore, a compound semiconductor device employing a compound semiconductor material with large electron mobility in the active region is generally being used in optical communication systems or wireless communication systems such as mobile phones, demanding high speed operation at a bandwidth of several dozen GHz.
On the other hand, the integration of a compound semiconductor device onto a silicon substrate is quite difficult; thus, in the conventional high speed communication system, the high frequency circuit that operates in a GHz band has to be separated from the signal processing part composed of a silicon integrated circuit.
It is known that a wide range of mixed crystals can be formed between silicon (Si) and germanium (Ge), and a high speed semiconductor device using a SiGe binary mixed crystal in the active layer is being proposed. In the SiGe binary mixed crystal, distortion occurs due to the difference in the atom radius between the silicon (Si) and germanium (Ge). However, as a result of the above distortion, the symmetry of the crystals forming the mixed crystal decreases and the electrons are prevented from scattering, thereby greatly increasing the mobility of the carriers. A high speed semiconductor device using the above SiGe binary mixed crystal can be integrated with other silicon (Si) semiconductor devices onto a common silicon (Si) substrate, and is therefore quite advantageous.
In the SiGe binary mixed crystal, the band gap decreases due to the substitution by germanium (Ge) in the silicon (Si) crystal; however, by doping the SiGe mixed crystal into a p-type dopant and using this as the base layer of the silicon (Si) bipolar transistor, a band discontinuity that prevents the entering of the minority carriers into the emitter domain can be formed on the valence band side in between the base and the emitter layers. As a result, the emitter injection efficiency can be improved and high speed response characteristics can be realized in the above SiGe heterobipolar transistor, as in the conventional compound semiconductor heterobipolar transistor.
FIG. 1A shows the structure of a heterobipolar transistor 10 using the conventional SiGe binary mixed crystal; and FIG. 1B shows the band structure of the heterobipolar transistor 10 of FIG. 1A.
As shown in FIG. 1A, the heterobipolar transistor 10 is formed on a silicon (Si) substrate 11 that has element isolation trenches 11A and an n+ type well 11B. On the n+ type well 11B, an n-type silicon (Si) collector layer 12 and a thin base layer 13 that is made of a p-type SiGe binary mixed crystal, are formed in this order. The collector layer 12 and the base layer 13 form a mesa structure, and on the base layer 13, an n+ type silicon (Si) emitter layer 14 is formed. Typically, the collector layer 12 and the emitter layer 14 are doped with phosphorous (P) or arsenic (As) to a carrier density of approximately 5×1017 cm−3 and 3×1020 cm−3, respectively. On the other hand, the base layer 13 is doped by boron (B) to a carrier density of approximately 5×1018 cm−3 or above. On the emitter layer 14, an emitter electrode 15 is formed; on the base layer 13, base electrodes 16 are formed; and on the n+ type well 11B, collector electrodes 17 are formed. Thus, in the structure of FIG. 1A, the n+ type well 11B forms a collector contact layer.
As shown in the band structure drawing of FIG. 1B, the germanium (Ge) concentration in the base layer 13 changes so that it increases in the direction from the interface of the base layer 13 and the emitter layer 14 to the interface of the base layer 13 and the collector layer 12. As a result, the conduction band Ec in the base layer 13 tilts towards the collector layer 12. By having the above-described tilted structure in the base layer 13, the electrons are sped up upon passing the base layer 13 through dispersion due to the drift electric field caused by this tilt of the conduction band Ec. As a result, the operation speed of the bipolar transistor 10 can be increased. The heterobipolar transistor using the above-described SiGe binary mixed crystal is disclosed, for example, in U.S. Pat. No. 5,353,912.
The heterobipolar transistor 10 of FIGS. 1A and 1B is formed on the silicon (Si) substrate using a well-known technique in the field of silicon (Si) integrated circuits. Therefore, it can be easily integrated with other information processing circuits including analog circuits.
However, in the heterobipolar transistor 10 of FIGS. 1A and 1B, boron (B) used in the doping process of the base layer 13 is easily dispersed into the neighboring collector layer 12 or the emitter layer 14. Thus, this heterobipolar transistor is unstable for thermal processing.
In response to the above problem, a technique of incorporating a small amount of C (carbon) as a dopant to the SiGe binary mixed crystal base layer 13 so as to prevent the dispersion of B (boron) into the neighboring collector layer 12 or the emitter layer 14 has been proposed in the conventional art (refer to: Lanzerotti, et al., Appl. Phys. Lett. 70(23), Jun. 9, 1997; Osten, H. J., et al., J. Vac. Sci. Technol. B16(3), May/June 1998, pp. 1750–1753).
Specifically, in the heterobipolar transistor controlled to have an ideal distribution of B (boron) in the base layer 13 by incorporating C (carbon) with a controlled distribution profile, the thickness of the base layer 13 can be minimized and the dopant concentration in the base layer 13 can be maximized. In this way, excellent operational characteristics can be expected.
However, in the heterobipolar transistor, controlled to have an ideal distribution of B (boron) in the SiGeC base layer 13, that has actually been created by the inventors of the present invention, the noise figure (NF) obtained according to formula 1, disclosed below, is degraded to a value of approximately 0.8–1.2 dB and particularly, the S/N ratio in the high frequency band including the GHz band is notably degraded.
                    NF        =                  1          ⁢                                                                r                b                ′                                            r                g                                                          ⁢                                    r              e                                      2              ⁢                              r                g                                              ⁢                                                                .                                                                            (                                                                        r                          g                                                +                                                  r                          b                          ′                                                +                                                  r                          e                                                                    )                                        2                                                        2                    ⁢                                          a                      0                                        ⁢                                          r                      e                                        ⁢                                          r                      g                                                                                  ⁢                              (                                                      f                                          1.2                      ⁢                                              f                        T                                                                              ⁢                                                                                1                                              h                        FE                                                                                                  ⁢                                                            I                      cb                                                              I                      E                                                                      )                                                                        (        1        )            (H. F. Cooke, Solid State Design, February 1963, pp. 37–42) Note that in the above formula 1, rg denotes the signal source impedance, rb′ denotes the base resistance, re denotes the emitter resistance, and a0 denotes the base ground current gain.
According to formula 1, the NF of the heterobipolar transistor can be reduced by decreasing the base resistance rb′ and the emitter resistance re. Thus, by controlling the distribution of carbon (C) in the base layer 13 and by controlling the distribution of boron (B) so that the density of boron (B) is maximized in the base layer 13 and the base resistance rb′ minimized, and also, so that the penetration of boron (B) into the emitter layer 14 is prevented and the emitter resistance minimized, the NF of the heterobipolar transistor can supposedly be reduced.
However, in the heterobipolar transistor with the distribution of boron (B) in the above-described SiGeC base layer 13 controlled to be in an ideal state, the NF is degraded to a value of approximately 0.8–1.2 dB and particularly, the S/N ratio in the high frequency band including the GHz band is notably degraded, as previously mentioned.
FIG. 2 shows the results of an analysis of the distribution of boron (B), carbon (C), and germanium (Ge) in the collector layer 12 and the base layer 13 of the heterobipolar transistor having the structures shown in FIGS. 1A and 1B, the analysis being made through SIMS (secondary ion mass spectrometry), wherein the n-type collector layer 12 and the base layer 13 doped into p-type layers by B (boron) are successively formed in this order through the use of the low pressure CVD technique.
In the state of the heterobipolar transistor shown in FIG. 2, that is, the state before the formation of the emitter layer 14 and right after the formation of the base layer 13, a SiGe mixed crystal region 13A, which does not include carbon (C) and boron (B), is formed at the lower portion of the base layer 13 around the boundary with the collector layer 12, and a SiGeC mixed crystal region 13B including carbon (C) and boron (B) is formed thereon. Also, in the region 13B, the concentration of carbon (C) is higher than the concentration of boron (B), and the distribution of boron (B) is limited to the range in which carbon (C) is distributed.
FIG. 3 shows the distribution of germanium (Ge) carbon (C), boron (B), and phosphorous (P), wherein the n-type emitter layer 14, which has been doped with P (phosphorous), is formed onto the structure of FIG. 2 using the low pressure CVD technique, after which a thermal process (700–800° C.) corresponding to the formation process of the CVD insulating film for the actual fabrication of the device is performed (refer to FIG. 11B, which is to be described later on), followed by a rapid thermal annealing (RTA) process at 900–1000° C. corresponding to the actual fabrication process of the device.
FIG. 12 is another diagram showing the distribution of the elements within the layers upon the creation of the heterobipolar transistor having the structure of FIG. 1. FIG. 2 shows the results from a SMIS analysis of the layers of the heterobipolar transistor in a state where each of the layers is deposited. The analyzing capabilities of FIG. 2 in a direction along the depth of the layers is somewhat limited. Thus, FIG. 12 shows a more accurate analysis of the distribution of the elements within the layers.
With reference to FIG. 3 and FIG. 12, the low pressure CVD process upon the formation of the insulating film is accompanied by the dispersion of C (carbon) in the base layer 13B into the entire base layer 13 so that the distinction between the regions 13A and 13B of FIG. 2 disappears so as to form a single base layer 13. As a result, the B (boron) is also dispersed in comparison with the state in FIG. 2; however, its distribution profile corresponds to that of carbon (C) and is precisely controlled so that boron (B) will not enter the emitter layer 14. Also, as shown in FIG. 12, a concentration gradient of germanium (Ge) does not exist on the collector layer 12 side of the base layer 13, and the germanium (Ge) concentration suddenly increases at the interface of the collector layer 12 and the base layer 13 from 0% to 16%, for example. Additionally, as indicated in FIG. 12, the germanium (Ge) concentration gradient on the emitter layer 14 side of the base layer 13 is not fixed, and the germanium (Ge) concentration suddenly decreases around the interface of the base layer 13 and the emitter layer 14 from 5–10% to 0%. Further, as indicated in FIG. 3, there is hardly any difference between the distribution profile of germanium (Ge) in the base layer 13 before and after the formation of the emitter layer 14.
However, the above distribution of carbon (C) prevents an ideal distribution of boron (B). In the heterobipolar transistor having a structure in which boron (B) is trapped inside the base layer 13, the desired characteristics cannot be obtained as previously mentioned.